Posts by davidryderuk

Hi,

Just to report back on my most recent experience of testing the HD Star v3 device's compatibility with LibreELEC:

LibreELEC 9.2 + default drivers: No device in TVheadend (see here for log)
LibreELEC 9.2 + 'DVB drivers for TBS': Partially working, but lots of transponders don't work (see here for log)

LibreELEC 9.2 + 'DVB drivers for DigitalDevices': No device in TVheadend (see here for log)

LibreELEC 9.2 + 'DVB drivers from the latest kernel': No device in TVheadend (see here for log)

At the moment I've found the most reliable option is using the Montage-3103b.v2 branch from b-rad-NDi's media_tree repository, which is based on Linux version 5.2-rc4. Brad appears to have done a lot of work cleaning up the code required for support of the 3103b device. So far the work seems to have been restricted to supporting the 3103b demodulator itself and any associated Hauppauge devices, hence I needed to write a small patch for the dw2102 driver, to get it to use Brad's updated code when the HD Star v3 USB device is detected. I've posted the patch, and some additional information on the GitHub page for Brad's media_tree branch (see here). I've no idea if the patch meets any particular set of guidelines or standards published for the Linux kernel, and, as far as I am aware, it hasn't been tested by anyone else, but it is currently the most reliable way for me to use the satellite tuner, being more reliable than any of the LibreELEC drivers, or any approach I may have previously described in this thread.

Quote from CvH

maybe you should have a look here ?h=montage-3103b.v2

this is likely upstreamd from Brad and if you have some patch he is likely helpful in getting it upstream too

This is his gh repo, just open an issue if you have questions/patches for the driver Issues · b-rad-NDi/Ubuntu-media-tree-kernel-builder · GitHub

Thanks. I just found that yesterday. It looks really promising, and it will be great when the code gets pushed upstream. I've only really had time to install the pre-compiled kernel from his PPA, which doesn't seem to work yet, though I didn't spend long trying to get it to work, so I could have missed something obvious.

Quote from NukeMania

but still some of transponders are not showing or green/broken video

The upstream driver for the M88DS3103 de-modulator uses target_mclk values between 96000 and 192000, depending on the transport's symbol rate (see here). CrazyCat's driver appeared to behave the same way, until he introduced support for the M88DS3103B revision, where the maximum target_mclk value was reduced to 144000 (see here). Therefore I guess you could potentially use target_mclk values up to 192000, which may help, but in my particular case I couldn't really spot any difference.

Thanks for including your patch.

Quote from NukeMania

with ts_clk=16000 and target_mclk = 144000 copied from m88rs6000.c values

more channels shows up and running good but not perfect still some channels are not stable

Any chance the channels which still don't work are on transponders with a bit rate higher than 53 Mb/s? I'm using channels available from the UK at 28.2°E, so I haven't really been able to test it with any transponder carrying more data than that, unless you count the 8k test channel, which doesn't work. I might have another go with the upstream drivers at some point, and see if they are working any better.

Quote from Booza

Would love to implement your fix but not sure how I can go about it on Libreelec!?

I downloaded and compiled the driver myself, using these instructions. If you are quite new to Linux, and haven't compiled much code before, it may be easier to wait for further developments.

Quote from Booza

Only problem is that I can't watch certain BBC HD Channels as it produces continuity errors. I presume this is a driver problem from the research I've been doing

I had that issue as well. Can you access any of the ITV HD channels? I found a solution which appears to work. Maybe someone like CrazyCat could look at my solution, and see if the change is safe enough for other users to test? The only other issue I've had, which I think is unrelated, is occasionally, maybe every week or so, I'll get flooded with 'dw2102: i2c transfer failed.' messages and have to reboot. I'm not running a Raspberry Pi or any device which is particularly well supported by the Linux community, so I can't rule out the possibility my device just has slightly flaky USB drivers.

Quote from smp

Can anyone confirm that there is really an issue with FEC 3/4 transponders?

My DVBSky S960 died today and I am about to order Geniatech HD Star...

I think it might actually be the bitrate. The transponders which are being problematic all have a bitrate higher than around 48 Mbps (when using the calculations on en.kingofsat.net).

Hi Scummer,

If your device is from 2014 then it will be using an earlier version of the 'M88DS3103 demodulator' chip so it won't be exactly the same issue. If you have still got access to kernel 4.15.0-45 with the CrazyCat drivers installed I'd try enabling the ds3k module's debug option (i.e. create a conf file in the '/etc/modprobe.d' directory with the line 'options ds3k debug=1'). It won't fix the issue, but it may provide a few more details. If you can find out when the device stopped working you may be able to work out what changes broke support for your device. I'd also try the most recent stable Linux kernel version without Crazycat's drivers installed (i.e. deb packages available from here).

I sometimes get the 'dw2102: i2c transfer failed.' message. I think it has something to do with the USB connection being subject to a high load. Anyway, I'd try the following, if you haven't already:

- Double check the device isn't plugged into a USB 1 port (you can use the 'lsusb -t' command if you are unsure).

- Try not to plug the device into an external USB hub (at least until you've got it working correctly).

- Reset the S4600 device before starting the computer (i.e. unplug the power cable, wait a few seconds and then plug the power it back in).

Regards,

David

p.s.

You'd also probably get more attention if you started a new thread, since it is a different issue.

Sorry, I can only test it using symbol rates of 22000, 23000 and 27500 KSym/s. I have noticed a similar issue when transponders use Forward Error Correction of 3/4. I can tune into the transponder, but the feed is full of audio and visual distortions. My TV manages to tune into the same transponders without any issue, but it doesn't have any record function.

I managed to download the media-build source code off bitbucket and compile it against the most recent Ubuntu 16.04 kernel (4.15.0-45). My DVB-S2 device is now working correctly on an Intel Atom Media PC, which is great. Thanks very much to crazycat,@CvH and anyone else who helped get the device working.

Just for the sake of completeness, I double checked the LibreELEC build I reported on in my last post. The build seems to match the one uploaded to Index of /test/ds3103b/. The kernel log also confirmed modules were being added from the 'kernel-overlays/driver.dvb.crazycat' folder. As before, I couldn't get the device to load.

Quote from CvH

did you use the build from Index of /test/ds3103b/ ??

the RPi image is confirmed working

I'm glad to hear it works for other people. I'll go through and double check if I've missed anything (which is more than likely). Is there a log file for the working version?

Hi CvH,

I have a Geniatech HDStar device which I bought around March 2018. I've failed to get it working with Linux. I think it probably has the 3103B chip, but I'm not certain (see my earlier post here). I ran your build with the crazycat drivers and have attached the log. I couldn't see the adapter listed in TVHeadend, but I'm not very familiar with that software, since I normally use MythTV.

Thanks,

David

Hi,

I recently purchased a Geniatech HD Star device. Output from 'lsusb -v' is the same as that posted on linuxtv, however I've had some trouble getting it to work. I've tried compiling kernel 4.11 with the patch available here and mentioned here. I didn't have any issues compiling the kernel, but the device failed to load. I've also tried the Ubuntu Xenial 4.13 kernel and OpenELEC 8.2.5. In each case there has either been a null pointer error or 'no frontend was attached'. The device always seems to be in a 'warm state', so no firmware is loaded. The Windows driver shipped with the device is dated December 2017. The only thing which seems to work is the remote control. Is there any way I can check if this problem is due to the new 'M88DS3103B' demodulator chip? If so, is there any information I can provide which may help?

Thanks,

David